Molecular Characterization and Enological Potential of A High Lactic Acid-Producing Lachancea thermotolerans Vineyard Strain

Lactic acid production is an important feature of the yeast Lachancea thermotolerans that has gained increasing interest in winemaking. In particular, in light of climate change, the biological acidification and ethanol reduction by the use of selected yeast strains may counteract the effect of global warming in wines. Here, the enological potential of a high lactate-producing L. thermotolerans strain (P-HO1) in mixed fermentations with S. cerevisiae was examined. Among the different inoculation schemes evaluated, the most successful implantation of L. thermotolerans was accomplished by sequential inoculation of S. cerevisiae, i.e., at 1% vol. ethanol. P-HO1produced the highest levels of lactic acid ever recorded in mixed fermentations (10.4 g/L), increasing thereby the acidity and reducing ethanol by 1.6% vol. L. thermotolerans was also associated with increases in ethyl isobutyrate (strawberry aroma), free SO2, organoleptically perceived citric nuances and aftertaste. To start uncovering the molecular mechanisms of lactate biosynthesis in L. thermotolerans, the relative expressions of the three lactate dehydrogenase (LDH) paralogous genes, which encode the key enzyme for lactate biosynthesis, along with the alcohol dehydrogenase paralogs (ADHs) were determined. Present results point to the possible implication of LDH2, but not of other LDH or ADH genes, in the high production of lactic acid in certain strains at the expense of ethanol. Taken together, the important enological features of P-HO1 highlighted here, and potentially of other L. thermotolerans strains, indicate its great importance in modern winemaking, particularly in the light of the upcoming climate change and its consequences in the grape/wine system.


Introduction
Winemaking is principally conducted by the yeast Saccharomyces cerevisiae. However, several other yeast species originating from grape berry skins, known as non-Saccharomyces or wild yeasts, may evolve during grape must fermentation. Recent studies on the description of yeast community evolution during alcoholic fermentation have revealed several species and strains present mainly at the early-

Fermentations in Pasteurized Grape Must
Triplicate fermentations were carried out in a pasteurized grape must (70 • C, 10 min) of Vilana cv. from Peza with the following characteristics: sugars 200 g/L; pH 3.42; titratable acidity 5.46 g/L, as tartaric acid; yeast assimilable nitrogen 240 mg/L; total sulphur dioxide 30 mg/L. Fermentations were conducted at 20 • C in 1 L Erlenmeyer flasks containing 750 mL clarified must and supplied with fermentation locks to allow only CO 2 to escape. Yeast inocula were propagated in grape must at 28 • C for 18 h under shaking (225 rpm). Each yeast strain was added at 6 Log CFU/mL under the following inoculation protocols: indigenous S. cerevisiae PzV6 strain (Is); L. thermotolerans P-HO1 strain and PzV6 added simultaneously (SmLt); P-HO1 followed by PzV6 after ca. 1% vol. ethanol production (SqLt); and commercial S. cerevisiae (Cs). The fermentation course was followed by monitoring the CO 2 exhaust as measured by weight loss.

Fermentations in Natural Grape Must
Fermentations were conducted in triplicate at 20 • C, in 2.5 L food grade plastic vessels with 2.2 L clarified must of Vilana cv. (sugars 207.6 g/L; pH 3.31; titratable acidity 5.48 g/L, as tartaric acid; yeast assimilable nitrogen 240 mg/L; total sulphur dioxide 30 mg/L). The vessel was closed with a silicone stopper supplied with a muller valve containing glycerol 50% v/v to allow only CO 2 to escape. Yeast inocula were propagated in grape must at 28 • C for 18 h under shaking (225 rpm). L. thermotolerans P-HO1 and S. cerevisiae PzV6 were added at 6 log CFU/mL in Is, SmLt and SqLt inoculation schemes, as described in pasteurized must fermentations. Spontaneous fermentations (Sp) were also conducted as reference. The course of the fermentation was monitored by density measurements.

Microbiological Analysis
Yeast populations were estimated by plate counts, using Wallerstein laboratory nutrient agar (WL; Lab M Bury, Lancashire UK), lysine agar (LA; Lab M Bury, UK) and ethanol sulfite agar (ESA) for the enumerations of total yeasts, non-Saccharomyces yeasts and S. cerevisiae, respectively. Plates were incubated at 28 • C for 2-5 days. Putative L. thermotolerans and S. cerevisiae colonies isolated from the initial, middle and final stages of non-sterile fermentations were examined microscopically and subjected to genotyping as previously described [9,21].

Chemical Analysis
Total and volatile acidity, total and free SO 2 , and pH of musts and wines were determined according to the methods in the Compendium of International Methods of Analysis of Musts and Wines [22]. Yeast assimilable nitrogen (YAN) was estimated with the formol titration method as described by Gump et al. [23]. Wine organic acids (citric, tartaric, malic, succinic, lactic and acetic acid), sugars (glucose, fructose), glycerol and ethanol were determined on a JASCO HPLC system (JASCO International Co. Ltd., Tokyo, Japan), consisting of a quaternary gradient pump (PU-2089 plus), an auto sampler (AS-1555) and a refractive index detector (RI-930), as described previously [3]. The major volatile components (i.e., acetaldehyde, ethyl acetate, methanol, 1-propanol, 2-methyl-1-propanol (isobutanol), 3-and 2-methyl-1-butanol) of wines produced from pasteurized grape must were estimated by direct injection of wines in a gas chromatograph (GC 8000 series, model 8060, Fisons Instruments, Milan, Italy) equipped with a split/splitless injector and flame ionization detector, as described by Nisiotou et al. [3]. The volatile compounds of wines produced from natural grape must were determined using a headspace solid-phase microextraction gas chromatography-mass spectrometry (SPME/GC-MS) method [24] with slight modifications [3]. Peaks were quantified using internal standard (IS) calibration curves. If authentic compounds were not available, the concentrations were expressed relatively to IS, i.e., by dividing the peak area of the compound of interest by the peak area of the IS and multiplying this ratio by the concentration of the IS.

RNA Preparation and First-Strand cDNA Synthesis
Cells were grown in 50 mL filter-sterilized grape must at 28 • C in 250 mL Erlenmeyer flasks under shaking (225 rpm). Cells (2 mL) were collected at early stationary phase (cultivation for 16 h) and centrifuged at 14,000 rpm for 5 min. They were then washed in dd H 2 O, centrifuged again, and resuspended in 150 µL dd H 2 O. To this cellular suspension, 150 µL of a breaking buffer solution (4% v/v Triton X-100, 0.2% w/v SDS, 200 mM NaCl, 20 mM Tris-HCl pH 8.0, 2 mM EDTA) and 300 µL phenol were sequentially added. Eppendorf tubes were then vigorously vortexed for 5 min, incubated at 65 • C for 5 min, vortexed again for 3 min, and centrifuged (14,000 rpm, 8 min). The supernatant was transferred into a new tube, an equal volume of phenol/chloroform (1:1) was added, vortexed and centrifuged as above. The above phenol/chloroform extraction step was repeated solely with chloroform. The aqueous phase was then transferred into a new tube in which 500 µL ice-cold 100% ethanol were added and placed in deep freeze (−80 • C) for 30 min. After centrifugation for 25 min at 4 • C, the pellet was air-dried and diluted in 20 µL dd H 2 O. RNA quality and quantity were assessed using NanoDrop 2000 (Thermo Scientific, Wilmington, DE, USA) and agarose gel electrophoresis. First-strand cDNA was synthesized from about 1 µg of total RNA with a PrimeScript RT Reagent Kit that includes a genomic DNA elimination reaction (Perfect Real Time; Takara, Otsu, Japan), according to the manufacturer's protocol.

RT-qPCR Analysis
Gene expression analysis was conducted through the reverse transcription quantitative polymerase chain reaction (RT-qPCR). About 15 ng of cDNA was amplified using KAPA SYBR FAST qPCR Master Mix (2×) Kit Real-time RT-PCR (Kapa Biosystems, Woburn, MA, USA) in a StepOnePlus TM (Applied Biosystems, Carlsbad, CA, USA) Real-Time PCR Detection System, according to the manufacturer's protocol. Amplification conditions consisted of 95 • C for 3 min, followed by 40 cycles of 95 • C for 20 s, 57 • C for 20 s, and 72 • C for 10 s, before melting curve analysis at the end of the run. All reactions were performed in triplicate. The primers used were designed using Primer3web version 4.0 (http://primer3.ut.ee/; part of services provided by ELIXIR -European research infrastructure for biological information). The genes analyzed and their targeted oligonucleotide sequences used are shown in Table 1. TAF10 was employed as a control housekeeping gene based on the report of Teste et al. [25]. The strain L. thermotolerans JCM 19085 (=CBS 6340; type strain) was used as a reference strain. The relative gene expression levels were calculated using the 2 −∆∆Ct method.

Sensory Analysis
Twelve experienced assessors (three male and nine female), members of the Institute of Technology of Agricultural Products, Athens, Greece, were recruited. Eleven aroma description terms (estery, apple/pear, orange, lemon, banana, melon, pineapple, floral, citrus, rose and jasmine) and seven taste/flavor terms (fruity flavor, acidity, sweetness, bitterness, viscosity, hotness and after-taste) were developed by the assessors in preliminary sessions. Wines were presented to panelists in a randomized order in duplicate. Each descriptor was scored by marking the perceived intensity on an unstructured 10 cm scale from 0 (absence) to 10 (high intensity).

Statistical Analysis
The chemical and sensory profiles of wines were compared by analysis of variance (ANOVA) and post-hoc Tukey's HSD test. Chemical parameters were analyzed by principal component analysis (PCA). Different inoculation protocols were compared by permutational multivariate analysis of variance (PERMANOVA). Statistical analyses were performed with PAST software version 3.11 [26] or SPSS v26 (IBM Corp., Armonk, NY, USA).

Fermentation Kinetics and Yeast Population Dynamics in Pasteurized Grape Must
Different inoculation scenarios were examined in pasteurized grape must fermentations, i.e., single inoculations with the indigenous strain PzV6 (Is) or the commercial S. cerevisiae strain Flavour 2000 (Cs), and mixed inoculations with L. thermotolerans P-HO1 and S. cerevisiae PzV6, added either simultaneously (SmLt) or sequentially (SqLt). Although PzV6 showed a more vigorous start up to 3.5 days in, it was then outreached by the commercial strain, which finished the fermentation 1.5 days earlier. The two strains also differed significantly in the total amount of CO 2 released (p < 0.001), being 94.45 g for the commercial strain and 91.35 g for PzV6. Almost identical kinetics, with no differences in the amount of CO 2 released, were observed for the Is and SmLt inoculation schemes, with SmLt being slightly slower after day 3. On the other hand, SqLt fermentations lasted longer than other fermentations by 2-3 days, also producing significantly lower levels of CO 2 (87.80 g; p < 0.001).
Both S. cerevisiae strains showed similar dynamics in single inoculations, reaching maximum populations of 8.20 ± 0.07 and 8.05 ± 0.56 Log CFU/mL for the indigenous and commercial strains, respectively. The strain ScMM23 exhibited identical kinetics in Is and SmLt fermentations ( Figure 1). However, PzV6 reached lower population density by ca. 0.67 Log CFU/mL in SqLt compared to Is, which reached maximum populations of 7.53 ± 0.04 Log CFU/mL. On the contrary, P-HO1 counts were higher in SqLt than in SmLt fermentations by 0.62 Log CFU/mL (7.90 ± 0.33 vs. 7.27 ± 0.04 Log CFU/mL, respectively).

Fermentation Kinetics and Yeast Population Dynamics in Natural Grape Must
Is and SmLt exhibited the highest fermentation rates compared to other treatments. Sp showed a rather long lag phase. SqLt had a more dynamic start compared to Sp, but both treatments showed the longest fermentation duration by ca. 3 days compared to all other treatments.
Single inoculations with either the commercial S. cerevisiae or PzV6 reached maximum cell densities of 8.05 Log CFU/mL. In SmLt, PzV6 showed similar kinetics as in Is, with a maximum population of 7.96 ± 0.06 Log CFU/mL (Figure 2a,b). Counts were lower in SqLt (7.50 ± 0.03 Log CFU/mL) and Sp (7.61 ± 0.05 Log CFU/mL) fermentations (Figure 2c,d) than in other treatments. The dominance of inoculated S. cerevisiae strains was confirmed by interdelta PCR fingerprinting at the middle and final stages of fermentations, except for Sp ferment.
Strain P-HO1 showed similar kinetic profiles in natural and pasteurized SqLt fermentations (maximum population of 7.74 ± 0.12 Log CFU/mL and 7.90 ± 0.33 Log CFU/mL, respectively). In SmLt inoculations, counts of P-HO1 were significantly lower in fermentations of natural must than in pasteurized must (maximum population of 6.06 ± 0.05 Log CFU/mL vs. 7.27 ± 0.04 Log CFU/mL, respectively). The addition of S. cerevisiae suppressed indigenous non-Saccharomyces (NS) yeast populations, which reached maximum populations of only 4.63 ± 0.01 Log CFU/mL and 3.48 ± 0.01 Log CFU/mL in Cs and Is, respectively ( Figure 2).

Fermentation Kinetics and Yeast Population Dynamics in Natural Grape Must
Is and SmLt exhibited the highest fermentation rates compared to other treatments. Sp showed a rather long lag phase. SqLt had a more dynamic start compared to Sp, but both treatments showed the longest fermentation duration by ca. 3 days compared to all other treatments.
Single inoculations with either the commercial S. cerevisiae or PzV6 reached maximum cell densities of 8.05 Log CFU/mL. In SmLt, PzV6 showed similar kinetics as in Is, with a maximum population of 7.96 ± 0.06 Log CFU/mL (Figure 2a,b). Counts were lower in SqLt (7.50 ± 0.03 Log CFU/mL) and Sp (7.61 ± 0.05 Log CFU/mL) fermentations (Figure 2c,d) than in other treatments. The dominance of inoculated S. cerevisiae strains was confirmed by interdelta PCR fingerprinting at the middle and final stages of fermentations, except for Sp ferment.
Strain P-HO1 showed similar kinetic profiles in natural and pasteurized SqLt fermentations (maximum population of 7.74 ± 0.12 Log CFU/mL and 7.90 ± 0.33 Log CFU/mL, respectively). In SmLt inoculations, counts of P-HO1 were significantly lower in fermentations of natural must than in pasteurized must (maximum population of 6.06 ± 0.05 Log CFU/mL vs. 7.27 ± 0.04 Log CFU/mL, respectively). The addition of S. cerevisiae suppressed indigenous non-Saccharomyces (NS) yeast populations, which reached maximum populations of only 4.63 ± 0.01 Log CFU/mL and 3.48 ± 0.01 Log CFU/mL in Cs and Is, respectively ( Figure 2).

The Effect of Different Inoculation Schemes on the Chemical Profiles of Wines Produced from Pasteurized Grape Must
The chemical characteristics and major volatiles of ferments from pasteurized must are shown in Tables 2 and 3, respectively. As shown by PERMANOVA, the chemical profile of ferments was significantly affected by the inoculation protocol applied (F = 121.2, p = 0.0001). Principal Component Analysis (PCA) was applied to differentiate among the chemical profiles of different ferments ( Figure  3). The first two principal components (PC1 and PC2) described 54.4% and 32.3% of the total variability, respectively. Is-and SmLt-derived profiles were grouped closely together, exhibiting high values of malic acid and 3-methyl-1-butanol. Cs formed a distantly separated group along the PC1 and PC2 axes, showing high levels of ethanol and pH. SqLt was placed on the PC1 axis, exhibiting high negative scores for lactic acid and propanol. The means of each individual chemical parameter were compared among the ferments produced by different fermentation protocols. The use of L. thermotolerans significantly increased acidity levels. Volatile acidity, lactic acid, acetic acid, ethyl acetate, propanol and acetaldehyde were significantly higher in SqLt compared to other treatments. On the other hand, the ethanol content was significantly reduced. Isobutanol and 3-methyl-1-butanol were increased in Is ferment, while 2-methyl-1-butanol was increased in Cs.

The Effect of Different Inoculation Schemes on the Chemical Profiles of Wines Produced from Pasteurized Grape Must
The chemical characteristics and major volatiles of ferments from pasteurized must are shown in Tables 2 and 3, respectively. As shown by PERMANOVA, the chemical profile of ferments was significantly affected by the inoculation protocol applied (F = 121.2, p = 0.0001). Principal Component Analysis (PCA) was applied to differentiate among the chemical profiles of different ferments (Figure 3). The first two principal components (PC1 and PC2) described 54.4% and 32.3% of the total variability, respectively. Is-and SmLt-derived profiles were grouped closely together, exhibiting high values of malic acid and 3-methyl-1-butanol. Cs formed a distantly separated group along the PC1 and PC2 axes, showing high levels of ethanol and pH. SqLt was placed on the PC1 axis, exhibiting high negative scores for lactic acid and propanol. The means of each individual chemical parameter were compared among the ferments produced by different fermentation protocols. The use of L. thermotolerans significantly increased acidity levels. Volatile acidity, lactic acid, acetic acid, ethyl acetate, propanol and acetaldehyde were significantly higher in SqLt compared to other treatments. On the other hand, the ethanol content was significantly reduced. Isobutanol and 3-methyl-1-butanol were increased in Is ferment, while 2-methyl-1-butanol was increased in Cs. Table 2. Chemical characteristics of wines produced from pasteurized grape must (mean ± SD, n = 3). Values with different superscript letters within each row differ significantly (p < 0.05).

Chemical Component
Inoculation Protocol 1

Cs Is SmLt SqLt
Total acidity (as tartaric acid g/L)

The Effect of Different Inoculation Schemes on the Chemical Profiles of Wines Produced from Natural Grape Must
The chemical characteristics and volatile metabolites of wines produced in natural musts are presented in Tables 4 and 5, respectively. PERMANOVA revealed significant discrepancies in the chemical profiles of wines produced through different inoculation protocols (F = 24.2, p = 0.0001). Differences in the chemical profiles were visualized by PCA (Figure 4). PC1 and PC2 accounted for 64.5% and 14.9%, respectively, of total variability. SqLt was negatively differentiated on PC2 for several attributes, such as total acidity, lactic acid and phenylethyl alcohol. Both Cs and Is were placed on the positive quadrant, showing highly positive scores on PC1 for 3-methylbutyl acetate, 2-phenylethyl acetate and hexyl acetate. SmLt was placed between Is and SqLt, while Sp was the most distantly located group among all ferments. Table 4. Chemical characteristics of wines produced from natural grape must (mean ± SD, n = 3). Values with different superscript letters within each row differ significantly (p < 0.05).

Chemical Component
Inoculation Protocol 1

.5. Transcriptional Analysis of L. thermotolerans LDH and ADH Genes
In order to shed some light in the molecular mechanisms of lactate biosynthesis in hermotolerans, the relative expressions of three LDH genes encoding lactate dehydrogenase, the k nzyme implicated in lactate biosynthesis, were evaluated in vineyard-associated L. thermotoler trains producing high levels of lactate (strains P-HO1, P-PT4 and P-PHO8, producing 15.6 g/L, 1 /L and 10.7 g/L, respectively) or low levels of lactate (strains A1L7 and P-PT5 producing 0.6 g/L a .4 g/L, respectively). The relative gene expression was estimated in respect to the type strain hermotolerans, which produced low amounts of lactic acid (0.5 g/L). Substantial differences w bserved in the expression levels of the three LDHs ( Figure 5). The LDH1 gene was highly express n all vineyard-strains, while LDH3 was consistently expressed at rather low levels in all strai omparable to those of the type strain. It is thus plausible that the differential production of lac cid may not be controlled through the transcriptional regulation of these genes. In contrast, DH2 gene was up-regulated in the high lactate-producing strains, but not in the low lacta

Transcriptional Analysis of L. thermotolerans LDH and ADH Genes
In order to shed some light in the molecular mechanisms of lactate biosynthesis in L. thermotolerans, the relative expressions of three LDH genes encoding lactate dehydrogenase, the key enzyme implicated in lactate biosynthesis, were evaluated in vineyard-associated L. thermotolerans strains producing high levels of lactate (strains P-HO1, P-PT4 and P-PHO8, producing 15.6 g/L, 14.8 g/L and 10.7 g/L, respectively) or low levels of lactate (strains A1L7 and P-PT5 producing 0.6 g/L and 1.4 g/L, respectively). The relative gene expression was estimated in respect to the type strain L. thermotolerans, which produced low amounts of lactic acid (0.5 g/L). Substantial differences were observed in the expression levels of the three LDHs ( Figure 5). The LDH1 gene was highly expressed in all vineyard-strains, while LDH3 was consistently expressed at rather low levels in all strains, comparable to those of the type strain. It is thus plausible that the differential production of lactic acid may not be controlled through the transcriptional regulation of these genes. In contrast, the LDH2 gene was up-regulated in the high lactate-producing strains, but not in the low lactate-producing strains. In order to investigate whether the high production of lactate might be also related to down-regulation of alcohol dehydrogenase genes (ADHs), the enzyme responsible for converting acetaldehyde into ethanol during alcoholic fermentation, both the L. thermotolerans ADH genes were included in the comparative transcriptional analysis. As is shown in Figure 5, the expression levels of both ADH1 and ADH2 were similar to those of the type strain, and no substantial differences were observed between low lactate-producing and high lactate-producing strains. Taken together, LDH2 was the only gene in the present analysis that had transcript levels that coincided with the levels of lactic acid production in different strains.
acetaldehyde into ethanol during alcoholic fermentation, both the L. thermotolerans ADH genes were included in the comparative transcriptional analysis. As is shown in Figure 5, the expression levels of both ADH1 and ADH2 were similar to those of the type strain, and no substantial differences were observed between low lactate-producing and high lactate-producing strains. Taken together, LDH2 was the only gene in the present analysis that had transcript levels that coincided with the levels of lactic acid production in different strains.

Sensory Analysis
The average scores assigned by the panel for each attribute of Vilana cv. wines are shown in Figure 6. Six attributes were found to differ significantly (p < 0.05) among wines, i.e., apple/pear, pineapple, jasmine, fruit flavor, acidity and sweetness. Acidity was significantly higher in SqLt than other treatments, followed by SmLt. Sp showed significantly higher sweetness, fruit flavor and jasmine or pineapple aromas than other wines. Jasmine and pineapple aromas were also highly rated in SqLT and SmLt, respectively. Cs-derived wine was characterized by high apple/pear aroma. Compared to other treatments, SqLt was characterized by higher orange and lemon nuances, reduced hotness and good aftertaste.

Sensory Analysis
The average scores assigned by the panel for each attribute of Vilana cv. wines are shown in Figure 6. Six attributes were found to differ significantly (p < 0.05) among wines, i.e., apple/pear, pineapple, jasmine, fruit flavor, acidity and sweetness. Acidity was significantly higher in SqLt than other treatments, followed by SmLt. Sp showed significantly higher sweetness, fruit flavor and jasmine or pineapple aromas than other wines. Jasmine and pineapple aromas were also highly rated in SqLT and SmLt, respectively. Cs-derived wine was characterized by high apple/pear aroma. Compared to other treatments, SqLt was characterized by higher orange and lemon nuances, reduced hotness and good aftertaste.

Discussion
Lactic acid production is an important feature of L. thermotolerans metabolism that has gained increasing interest in enology in light of climate change. Global warming, in particular, is expected to have a significant impact on wine quality, by reducing acidity and increasing ethanol content.

Discussion
Lactic acid production is an important feature of L. thermotolerans metabolism that has gained increasing interest in enology in light of climate change. Global warming, in particular, is expected to have a significant impact on wine quality, by reducing acidity and increasing ethanol content. Therefore, biological acidification, and the development of ethanol reduction procedures by the use of appropriate yeast species/strains, could offer an advantageous alternative in modern winemaking, especially in warm viticultural regions. Despite the high importance of lactate biosynthesis in L. thermotolerans for winemaking and also for various other biotechnological applications, the performance of high lactic acid-producing strains in wine fermentation has not been studied in depth, and the respective molecular mechanism underlying the unusually high levels of lactic acid production in this species is largely unknown [13]. Accumulating data show that L. thermotolerans isolated from vineyards or other ecological niches may exhibit significantly different metabolic activities [9,13]. There is considerable diversity among strains in the production of various metabolites, including ethanol, acetic acid, esters, terpens and lactic acid or other secondary metabolites of alcoholic fermentations.
Here, a high lactate-producing L. thermotolerans strain (P-HO1) that had been previously identified [9] was further characterized, and its enological potential in mixed fermentations with S. cerevisiae was evaluated. As expected, in simultaneous inoculations the P-HO1 population was suppressed by S. cerevisiae, probably due to the highly antagonistic features of S. cerevisiae in grape must fermentations (e.g., high sugar fermentation capacity and nitrogen uptake) [27,28]. As a result, the metabolic contribution of P-HO1 in SmLt ferment was limited. Contrastingly, in sequential inoculations with S. cerevisiae (SqLt), P-HO1 was highly active and capable of producing high levels of lactic acid. Notably, lactic acid levels exceeded 10 g/L in SqLt, which, to our knowledge, is the highest amount ever recorded in mixed-inoculated alcoholic fermentations. Subsequently, the total acidity was increased by 9.2 g/L, which is also the highest rise ever observed. Ethanol content declined greatly, by 1.6% vol. in sterile and 0.6% vol. in natural grape must. Such reduction may counterbalance the average increase in alcohol content of wines in warmer regions, which has been estimated to be approximately 2% over the last 30 years [29].
Strain P-HO1 produced significant levels of lactic acid during fermentation of natural grape must, albeit lower than in pasteurized must. This might be due to the lower prevalence of L. thermotolerans in natural than in pasteurized must, since the production of lactic acid depends on the concentration of yeast cells [6]. In addition, L. thermotolerans has been shown to exhibit limited maximum population when low oxygen is present [8]. Taken together, the above features indicate that strain P-HO1 could serve as a promising alternative for efficient acidification and concurrent ethanol reduction in winemaking, offering to modern winemakers an alternative for producing balanced wines with non-invasive chemical or physicochemical means. P-HO1 was shown to lower the levels of esters in SqLt, in accordance with previous observations [8,[30][31][32]. However, here it was also noticed that the reduction was associated with increases in ethyl isobutyrate (strawberry aroma), as well as organoleptically perceived citric (lemon and orange) and jasmine nuances. The use of L. thermotolerans was also accompanied by some more interesting organoleptic characteristics, such as decreased hotness and good aftertaste. Another positive effect of acidification was the increase in the levels of free SO 2 , facilitating thereby a reduction in total SO 2 supplementation. Our data show that the successful implantation of L. thermotolerans is accomplished by late sequential inoculation of S. cerevisiae, i.e., after ca. 1% vol. ethanol production, corroborating previous results [6,8]. By this means the acidification of fermented must is expedited, since lactic acid is produced at the early stages of fermentation [5].
Although the production of lactic acid is a key metabolic characteristic of L. thermotolerans, there is substantial heterogeneity among strains in the amount of lactic acid they produce [9]. During fermentative sugar metabolism in L. thermotolerans, ethanol and lactate are the predominant products along with acetate, which is excreted in small amounts with minimal inter-strain variation [9,13]. Thus, the most important steps in pyruvate metabolism of L. thermotolerans during grape must fermentation seem to be the conversion of pyruvate into lactate by the activity of lactate dehydrogenase (LDH), and to ethanol by the function of alcohol dehydrogenase (ADH) (Figure 7). Based on the whole genome sequence of the type strain [33], L. thermotolerans possesses three LDH (LDH1, LDH2 and LDH3) and two ADH (ADH1 and ADH2) paralogous genes. Here we asked whether the high production of lactate in strain P-HO1, and other strains as well, might be attributed to up-regulation of LDH and/or down-regulation of ADHs genes. Comparative transcriptional analysis in high and low lactate-producing strains revealed a potential role of LDH2, but not of the other genes. Another interesting observation made was the differential regulation of the three LDH paralogs. It would be important to see whether the expression differences of these genes are coupled with functional differences, as in the case of the S. cerevisiae ADH genes [34,35]. fermentative sugar metabolism in L. thermotolerans, ethanol and lactate are the predominant products along with acetate, which is excreted in small amounts with minimal inter-strain variation [9,13]. Thus, the most important steps in pyruvate metabolism of L. thermotolerans during grape must fermentation seem to be the conversion of pyruvate into lactate by the activity of lactate dehydrogenase (LDH), and to ethanol by the function of alcohol dehydrogenase (ADH) (Figure 7). Based on the whole genome sequence of the type strain [33], L. thermotolerans possesses three LDH (LDH1, LDH2 and LDH3) and two ADH (ADH1 and ADH2) paralogous genes. Here we asked whether the high production of lactate in strain P-HO1, and other strains as well, might be attributed to up-regulation of LDH and/or down-regulation of ADHs genes. Comparative transcriptional analysis in high and low lactate-producing strains revealed a potential role of LDH2, but not of the other genes. Another interesting observation made was the differential regulation of the three LDH paralogs. It would be important to see whether the expression differences of these genes are coupled with functional differences, as in the case of the S. cerevisiae ADH genes [34,35]. Present results show that the use of the vineyard L. thermotolerans P-HO1 strain in mixed fermentations with S. cerevisiae PzV6, especially when inoculated in a sequential mode, may serve as a powerful and useful tool for reducing the alcohol content of wines through the production of lactic acid, thus leading to simultaneous biological acidification. The molecular mechanism underlying the unusually high production of lactic acid indicates the possible involvement of LDH2, rather than other LDH paralogs or ADH genes. Apart from several other biotechnological applications, the important features of P-HO1 and potentially of other L. thermotolerans strains are expected to be of Present results show that the use of the vineyard L. thermotolerans P-HO1 strain in mixed fermentations with S. cerevisiae PzV6, especially when inoculated in a sequential mode, may serve as a powerful and useful tool for reducing the alcohol content of wines through the production of lactic acid, thus leading to simultaneous biological acidification. The molecular mechanism underlying the unusually high production of lactic acid indicates the possible involvement of LDH2, rather than other LDH paralogs or ADH genes. Apart from several other biotechnological applications, the important features of P-HO1 and potentially of other L. thermotolerans strains are expected to be of high value for the wine industry. Based on the present results, future research on the genetic manipulation of yeasts for the overproduction of lactic acid could also consider the use of selected L. thermotolerans strains, or the heterologous expression of the L. thermotolerans LDH2 gene.